Churn Prediction Of High-end Clients In Commercial Banks

This Article examines how to predict the attrition of high-end clients in commercial banks using both qualitative and quantitative analysis. Based on the customer database from a Chinese commercial bank, we identify the attributes that significantly affect customer churn and then develop a prediction model using Cox's proportional hazard model.The article analyzes various factors that may cause customer attrition, examines their correlations with churn behavior, and illustrates the problems with univariable analysis. Then, significant predictors are selected by multivariable analysis, a customer churn prediction model is built using Cox's proportional hazard model, after checking the basic assumption of proportional hazards. At last, the model is valuated using the hold-out sample.The empirical results suggest that the significant prediction variables include: whether the customer is local resident, has saving accounts, uses bank cards, has closed at least one saving account or debit card, how many accounts has the customer have, how many deals have be done during the observation period, and when did the last deal happen. It also suggests that the proposed churn prediction model using these predictors exhibits satisfactory performance.